Don't you hate those boilerplate replies? It's a BLUE SCREEN. Kernel panic. Happens to me with AQtime and "User Time" or "User+Kernel Time" counters on Win7/64 *all the time*. And there is no way to predict when it's going to happen or recall what I was doing. Yesterday, for instance, it happend *four times*, one of them I wasn't even running the application, just sitting there browsing the results in AQtime.
Don't get me wrong, AQtime is a great product with very useful features, like the triggers, or selective profiling of only the loaded modules (VTune used to do that, where is it now?) but why can't they make it stable enough to be useful on a multicore multiprocessor machine? Elapsed time counters are almost meritless; I get 20-30% variation in results from two consecutive runs of my app with everything else the same, if I use the Elapsed Time counters, and I'm forced to use those because if I try the User Time, I get bluescreens. Random bluescreens. Those cost me not only data loss (yes, frequent project saves do help), but I also lose time because it takes several minutes for my machine to reboot, start AQtime again, and load a project... (any improvement expected in the instrumentation process? I have 23 cores sitting there doing nothing when your single-threaded instrumentation is chugging along processing a large complex DLL of ours...)